[01:06:29] does anyone here use AD integration? i cant for the life of me get it to work correctly [02:06:39] anyone around? [02:07:46] yes [02:07:47] !ask [02:07:47] Please feel free to ask your question: if anybody who knows the answer is around, they will surely reply. Don't ask for help or for attention before actually asking your question, that's just a waste of time – both yours and everybody else's. :) [02:08:24] oi my bad :/ [02:08:38] just trying to get AD integration working, and its making my brain melt lol [02:08:54] uh oh. that does sound unpleasant [02:11:06] i mean it looks like it works, but i just get a generic "database query error" after i try logging in with an AD user/password [02:27:25] is there a way to feed wikitext directly to a parse action? in other words, action=parse also wants a text=“” parameter, where the content for text can be wikitext. but I want to feed it the wikitext of a page, not have to include wikitext directly in the query like that [02:31:06] oh, nevermind, i see the page= parameter for parse [02:39:03] i’ve been digging around today to try and find a way to extract citation information from an arbitrary wikipedia article. I’m not finding any method for doing this other than: 1) grab the full wikitext of a page, and then 2) write my own parser to extract … pairs, or to focus on {{Cite … }} pairs. But however I’d write #2, I can’t find any way to make this easier with the API. No way to narrow what I have to [02:39:04] look at (say, to a references section), no way to automatically extract citations, etc… Am i correct on this? [02:55:46] Is the Phabricator update finished? I'm asking, because Phabricator currently does not allow creation of a new task. Is that a known issue? [02:56:36] Hi Joergi. [02:56:46] Joergi: I'm having the same issue. [02:56:56] Joergi: Can you please leave a note on https://phabricator.wikimedia.org/T127271 if that works for you? [02:58:41] Leah: Adding notes obviously is working. [02:58:51] Creating new tasks is not. [02:59:35] Obviously! :-) [02:59:51] Thanks for the note. Maybe there will be a pattern to see between our accounts. [02:59:59] I'm told the issue is being investigated already. [03:01:20] Mine is just an ordinary account. No special rights, no memberships in any groups (if something like that exists in Phabricator). [03:01:27] New login, no old cookies there. [03:01:59] Never mind, I'll switch to good old text files until this got fixed. [03:02:09] :-) [03:02:14] It should hopefully be fixed soon. [03:04:08] I just tried again and now it worked. [03:05:20] anyone running mediawiki with AD integration? i keep getting a stupid database query error [03:08:25] What's the error? [03:08:35] N9X: ^ [03:08:41] "A database query error has occurred. This may indicate a bug in the software." [03:08:55] its terribly specific i know :p [03:09:11] !debug [03:09:11] For information on debugging (including viewing errors), see http://www.mediawiki.org/wiki/Manual:How_to_debug . A list of related configuration variables is at https://www.mediawiki.org/wiki/Manual:Configuration_settings#Debug.2Flogging [03:09:21] Can you enable debugging mode to get the real error message? [03:09:35] yeah i can give that a shot [03:12:46] hmmm mediawiki doesnt seem to be working for me [03:13:43] mediawiki.org i mean [03:14:48] Not working how? [03:14:51] https://www.mediawiki.org/wiki/Manual:How_to_debug loads for me. [03:15:30] appears to be a DNS issue Xd [03:16:49] Guess you're just having bad luck. [03:19:11] yeah no kidding, go tthe site up now, trying to setup debugging [03:32:39] i cant even get logging to show me anything [03:46:57] does anyone else use LDAP integration with active directory? are there database settings that need to be setup in order for it to function correctly? [04:23:20] does anyone else use LDAP integration with active directory? are there database settings that need to be setup in order for it to function correctly? [04:30:43] well at least i got debugging to work, now i have a real error message [10:50:52] Will {{switch: }} work to switch background-color in a tag? [11:34:59] I don't see why not [11:35:30] [14:22:25] wathup my dawgs [14:33:49] anyone around who knows if markup works with the Labeled Section Transclusion extension (on Meta)? [15:31:44] Weirdness. We have a bunch of edits on our instance attributed to a user who isn't registered, and none of us can remember ever creating or deleting a user by that name. [15:32:29] And we have anonymous edits disabled, and account creation disabled (manual only, by admins) because we had big problems with vandalism. [15:33:13] These edits in question aren't vandalism, they're just unexpected -- how did someone who isn't registered get to make them? Is there some other login mechanism we're not aware of? [15:34:37] Could the account have been created before you cracked down on account creation? And just been dormant for a while? [15:35:56] Nah, we locked that stuff down in 2009/2010, and the edits in question are from 2011/2012... let me check those facts. [15:36:49] Yeah, all the spam vigilance was 2010, we had a lid on it by then. [15:38:05] An account has to exist to be dormant; as far as I can tell, this user never existed. There's no creation, deletion, block, or any other activity with their username in the whole log. [15:38:37] myself: There are ways to permanently remove a user account from the database. [15:38:39] that's what's baffling me. I'd expect these edits to show up attributed to an IP address, except, we have anonymous edits disabled. [15:39:00] myself: Dealing with QuickBooks spam? [15:39:05] Check the database to see when they were created [15:39:12] they weren't created! [15:39:18] https://www.i3detroit.org/wiki/Special:Contributions/MSGJ [15:39:23] Maybe during the spam cleanup, a wrong user account got deleted? [15:39:24] it's not spam, it's heplful edits, good stuff [15:39:44] just, good edits inexplicably attributed to a user who, as far as I can tell, doesn't exist. [15:39:49] I understand the problem. [15:40:07] What do you want to happen now? Do you want to add that account (again)? [15:40:13] I mean check the "user" (ow "mwuser") table directly [15:40:27] No, I want to understand the mechanism by which this happens, to understand if perhaps we failed to lock something down [15:40:50] cuz if this helpful person came in through this route, then someone malicious could too. [15:40:52] I don't think there is another way to create an account, to answer the original question. [15:41:26] As far as I can tell, at least with a normal MW installation, no strange extensions and so on, it is just impossible to create edits, which look as if they have been done by one user, although in fact they haven't. [15:41:26] Is there some cross-instance way that a user of one wiki could carry their credentials to another, without having an account there? [15:41:54] E.g. from a development system, if it uses the same DB, but a different user table. [15:42:11] Very strange. [15:42:50] However, since these edits are _not_ spam and you say you did extensive spam cleanup, I really think, that this row from the user table got deleted by accident. [15:42:51] I'm not aware of our tables ever having been switched around, huh. I'll ask the admins just to b esure. [15:43:39] In the revision table, when you look at one of the revisions, which this user has made, you will not only see his rev_user_text, but also the old user id [15:43:48] The id is in rev_user. [15:44:00] I don't think we ever deleted anyone, though. The spam accounts are all still there (I just checked Special:ListUsers) and just in (blocked) state. [15:44:19] Anyway, this row obviously is missing. [15:44:21] I'll run this by our admins, thanks. So puzzled. [15:44:28] With the ID and the name, you can add the missing row to the user table again. [15:45:13] I'm still curious to know if this is the QuickBooks attack. [15:45:45] JordiGH: What's that mean? We don't use quickbooks or have any related extensions loaded.. [15:46:11] I linked the contributions page above, you can see the edits in question, none of them are QB spam. None of them are spam at all. [15:46:15] myself: The content of the spam. A lot of Mediawiki instances have been targetted over the past few days. Looks like they're mechanicalturking the captchas. [15:46:57] myself: I didn't follow this from the beginning. You were talking about spam, so I assumed there was some other spam content elsewhere. Anyways, never mind, I guess you have your own problems. [15:47:03] No, no, no. It's not spam. It's not in the last few days. It's from 2012 and we just noticed it because some of the pages have red links, and we're trying to clean those up. [15:47:19] JordiGH: Are we speaking about spam with "support phone number", "tech support number", "service number" and so on? [15:47:22] It's only weird because the changes are attributed to a user whose contributions page says they don't exist. [15:47:25] Joergi: Yes. [15:47:53] JordiGH: Seen that. What do you think they are doing with the captchas? [15:48:26] "mechanical turk" is an Amazon service where you can pay real humans to do tasks that computers are bad at, such as solving captchas. [15:48:53] Ahhh. [15:49:08] myself: It's actually an older trick into pretending a machine is human by having an actual human do the work. Amazon took the name from a contraption called The Turk. [15:49:09] I don't think that is very satisfying. [15:49:28] I'm aware of the original chess machine. :) [15:49:35] How much are they paying? [15:49:48] I mean, who would spend his day solving captchas? [15:49:53] Joergi: QuasAtWork yesterday suggested that this spam attack was being worked like that. I don't think it has anything to do with Amazon. [15:50:01] Someone who lives in a low-wage part of the world. [15:50:09] Joergi: Impoverished Indians who have been promised some sort of profit if the scam succeeds. [15:50:22] myself: Obviously. [15:51:04] They have created a few domains, which then link to the spammed pages inside the wikis. [15:51:14] At least that is, what I see in Google Webmaster tools. [15:51:34] these spammers are actively adapting to everything we throw at them [15:51:48] they may be using bots to do the dirty work but they've got some brains behind the operation [15:51:52] Usually such spam attacks don't last. I still think what they do is stupid. [15:52:01] QuasAtWork: Adapting, oh yeah. [15:52:03] I just had to adjust my blacklists and abusefilter rule this morning, again. [15:52:31] Same for me. I think you know regexps...that's what you should use... [15:52:47] It is ... interesting to see, how they try until they got around... [15:52:48] if you look up quickbooks247.com the domain is registered to an Indian guy [15:53:01] the company claims to be located in Chicago and does have one IP that geolocates there. [15:53:17] I've also had attacks come from the company's nameservers [15:53:26] I just finished watching Mr Robot, so now I feel all leet-hackery. >_> [15:53:27] <_< [15:53:33] I'm sure that's a breach of terms of service there at least [15:53:43] Hm. Is that the spam with "tech support phone number" and so in it? [15:53:48] yeah [15:54:02] In my case, there were no links. Only phone numbers. [15:54:11] yup [15:54:12] Yeah. [15:54:20] but I traced the numbers on several of them [15:54:26] via toll free # managers [15:54:35] The scam works by having people call the phone number and then being "warned" that they are at risk of being audited for tax fraud. [15:54:37] the numbers belong to these same companies the spam is coming from [15:54:42] all registered to Indian names [15:54:45] It is a wide campain; when you google these numbers, you will also find videos on YouTub, which have this number in them. [15:55:04] Yes, obviously they expect to make a lot of money out of the scam. [15:55:49] QuasAtWork: But it sonds to me like you have a plan about wht can be done against this? [15:55:59] here's a good blacklist rule to add: .*[@]{2,}.* [15:56:13] almost no legitimate reason to ever have @@ (or @@@ etc) in an article title. [15:56:42] there's no instance of '@@' in my entire wiki's text :P [15:57:07] I bet the scammers are here and adjusting their attack as we speak. :P [15:57:08] Joergi: nah [15:57:18] I just wanted to understand the enemy so to speak so I did a bunch of digging [15:57:50] I'm sure you could report it to the FBI but it'll probably get put in File 13 [15:58:23] You mean the round one, next to the table? [15:58:45] right next to a disposable coffee cup, and a slowly browning banana peel [15:58:51] ;) [15:59:19] QuasAtWork: But you know, which company hosts the sites. If this is a known host from the US (youspoke of Chicago), maybe _they_ would be interested. [15:59:33] the company *is* the spammers [15:59:40] they all trace back to the same India-based operation [16:00:12] if you mean where their stuff is hosted I'd have to look that up again. [16:00:33] Exactly, I mean the host. [16:02:29] their nameservers are hosted by Softlayer Technologies Inc according to the IP [16:02:58] 4849 Alpha Rd. Dallas, TX, 75244, US +1-214-442-0601 www.softlayer.com [16:03:06] drop them a line at abuse@ [16:03:53] the quickbooks247.com resolves to the same IP as its nameservers [16:04:08] Their abse team has this email address: abuse@softlayer.com [16:04:09] a bit odd in and of itself [16:04:46] problem is most hosts want server logs that "prove" what happened [16:05:48] Logs from whom? From the spammed site? [16:06:00] yeah. [16:06:30] I think you have those logs (given that you have control over the server...)? [16:16:42] how would I grep for a specific IP, I'm terrible at that crap [16:17:00] grep 127.0.0.1 somefile? [16:18:39] didn't output anything [16:18:44] probably in a previous log file... [16:19:24] aaaand my admin has the rotated logs set to .gz themselves >_< [16:19:37] zgrep [16:20:16] zgrep 260.260.0.1 thelog thelogs*.gz [16:23:27] gzip: /var/log/apache2/access.log*.gz: No such file or directory [16:23:46] blatantly untrue. [16:25:31] I haven't seen apache logs in a while. What do the file names look like? [16:25:54] -rw-r----- 1 root adm 32M Feb 18 10:15 access.log [16:25:55] -rw-r----- 1 root adm 67M Feb 14 06:25 access.log.1 [16:25:55] -rw-r----- 1 root adm 1.9M Dec 13 06:25 access.log.10.gz [16:25:56] etc [16:27:13] maybe sudo isn't giving me good enough perms to do this. [16:29:30] Oh, I see the problem. [16:29:44] the thing is that the glob fails to get expanded before sudo ever sees it. [16:30:27] A non-privileged bash tries to expand glob, since it doesn't have permissions to see the log directory it just returns the unexpanded glob, and then sudo passes that unexpanded glob to zgrep and zgrep is like "wtf m8?" [16:30:44] QuasAtWork: Can you temporarily run `sudo -i` and then do this again? [16:35:40] where are user home directories on Debian [16:35:52] I am not used to being root [16:35:54] They're usually found in /home [16:35:58] so I cannot just do ~ [16:36:00] k [16:36:04] Might want to check /hotel [16:36:06] JUST IN CASE [16:36:48] grep worked under sudo -i but it was found in an other_vhosts_access log rather than the main one [16:37:22] Sorry, couldn't help myself. In other news, is there a fancy way I can make resource loader tell me the hostname or IP address of the frontend that's feeding it data? I've been using cookies up until now and it's become something of a problem [16:45:02] http://pastie.org/private/rbficnznayzymnxwjvsma cut down to the relevant portions (no load.php) [17:10:41] Composer update giving errros about SMW -> Remove SMW -> No more errors. Perfect. [17:10:45] *errors [17:20:38] Trela: That seems like cheating to me >> [17:21:20] Ulfr: Is it? I can not convince the SMW maintainers to fix their broken composer setup, so that is the work around. [17:22:38] Nooo idea, as far as I'm concerned composer is magic. And like the book says, never make a wizard angry [17:23:34] Composer works... some times. Kind of. [17:23:59] * Ulfr shrugs [17:24:09] I haven't had a problem with it, but I'm boring and keep to actual releases [17:24:13] Also that bit where I have to use nasty low level terminal commands to disable IPv6 because the composer.org people do not know how to fix their IPv6 DNS. [17:24:45] Yeah, it is a necessary evil for our setup here. [17:38:43] QuasAtWork: Looks good so far. [17:40:31] I reported it but we'll see what good it does. [17:41:02] Depending on the host this may or may not be enough. [17:41:15] I have seen some, who after such a report start investigating themselves. [17:41:26] In this case, results will be pretty clear... [19:48:58] hello I am new to mediawiki and I want to contribute.. I am following the instructions on this link https://www.mediawiki.org/wiki/Gerrit/Tutorial to set up the development environment but when I try doing this [dev@localhost ~]$ ssh -p -v 29418 dev24@gerrit.wikimedia.org it says 'Broken pipe' . Please help with this issue I appreciate it [19:50:20] GSoC season is upon us. [19:52:23] dev24_, what sort of network are you on? [19:52:45] Oh, Indian universities typically have restrictions on port 22. [19:52:57] I think we found some had 29418 blocked though, didn't we? [19:53:22] Anything but port 80, often. [19:53:55] Krenair: I do not understand the question completely however I am using Internet at home so I don't think a blocked port should be an issue [19:54:29] does it say broken pipe every time you run it? [19:55:53] Krenair, yes it does [19:57:18] dev24_, do you have any special ssh config? what does it show if you add -vvv instead of -v? [19:58:19] can you interspose flags like that? [19:58:26] e.g. -p -v [19:58:43] Hello! I need to reset my password but no e-mail is linked with my account. [19:58:49] that's probably a good question Skizzerz [19:59:02] JeffGarlin, uh... this is the MediaWiki channel. Is this about your mediawiki.org account? [19:59:04] if it's using getopt() I know that won't work [19:59:21] (or rather am 90% sure of that) [19:59:23] I'm talking about my wikipedia account. I am so confused as to how to get help to get back into my account. [19:59:33] -> #wikimedia-tech please [19:59:51] dev24_: try swapping the -p and -v, e.g. ssh -v -p 29418 ... [20:00:03] might work, might not, but easy enough to try :) [20:00:11] I'm sorry the -v was added at the end after i got an error when i typed '-p -v' i copied the wrong command here.. but with 'ssh -p 29418 dev24@gerrit.wikimedia.org -vvv' it says 'packet_write_wait: Connection to 208.80.154.81: Broken pipe' [20:00:22] ok [20:00:33] then disregard what I said [20:02:05] hmm [20:03:22] so https to gerrit works, but you get broken pipe on ssh only? [20:04:34] I wonder if the server is simply closing the socket instead of returning if key auth failed? either that or the issue is somewhere along the network [20:05:55] Krenair, Skizzerz : should i proceed to the next step where they clone 'examples' and see if that works for me? [20:06:13] In case it has authenticated the key.. [20:06:47] you could try, I doubt it'd work; more likely to fail with the exact same error [20:07:41] Skizzerz, no, I get "Permission denied (publickey)." if I use the wrong key [20:09:12] Krenair: if you succeed you get the gerrit page thingie? (not at the pc with my private key so can't check myself) [20:09:13] Should i just clone the repo anonymously? [20:09:27] cloning anonymously means you can't submit patches you make [20:09:40] (or well, you can, but you need to fix the remotes first) [20:09:52] Skizzerz: ok.. [20:10:12] Skizzerz, if I succeed, even I get "Hi Alex Monk, you have successfully connected over SSH. " etc. etc. etc. [20:10:26] ok so doesn't seem like a server misconfiguration then [20:10:31] cloning 'examples' worked! Does that mean i can clone the repo now? [20:10:35] You can submit patches over HTTPS [20:10:44] if you don't mind using the password system instead of keys [20:10:57] it's how I push from tin/mira to gerrit [20:34:11] i’ve been digging around today to try and find a way to extract citation information from an arbitrary wikipedia article. I’m not finding any method for doing this other than: 1) grab the full wikitext of a page, and then 2) write my own parser to extract … pairs, or to focus on {{Cite … }} pairs. But however I’d write #2, I can’t find any way to make this easier with the API. No way to narrow what I have to [20:34:11] look at (say, to a references section), no way to automatically extract citations, etc… Am i correct on this [20:36:25] mixwhit: you may get better luck parsing the HTML, finding
    [20:37:23] thought about that, though I appreciate and could use the meta data in the wikitext (e.g. last/first/etc). also, I was thinking the wikitext may be more stable over time than the CSS in the HTML?